Patient motion is a major issue in nuclear medicine imaging studies due to the substantial time to acquire enough counts to produce clinical-quality images. Motion correction algorithms exist but have limited accuracy and effectiveness, and in many instances the only practical alternative is to re-scan the patient. Most work on patient motion sensing has focused on optical tracking systems. Although sub-millimeter accuracy has been attained in research laboratories, the cost, complexity, and line-of-sight requirement of external optical tracking systems have prevented their routine use in clinical nuclear medicine departments. There is an unmet clinical need for a practical and inexpensive patient motion sensing device for nuclear medicine studies. This is especially true for brain PET and SPECT studies which demand high resolution and accurate quantification, both of which may be severely degraded by patient motion. We propose a novel and inexpensive method for head tracking for brain PET and SPECT studies that satisfies these practical requirements. Our aims in phase I are to develop the necessary hardware and software for accurate head motion sensing and to validate this method in clinical brain PET and SPECT studies.